Vol. 38 2012 No. 1
IZABELA SÓWKA*, ANNA ZWOŹDZIAK*, KRYSTYNA TRZEPLA-NABAGLO**, MARIA SKRĘTOWICZ*, JERZY ZWOŹDZIAK*
PM2.5 ELEMENTAL COMPOSITION
AND SOURCE APPORTIONMENT IN A RESIDENTIAL AREA
OF WROCŁAW, POLAND
Ambient PM2.5 aerosol samples were collected from Wroclaw residential area between January and April 2009, and their elemental compositions were studied. The mean mass concentration of PM2.5 was 36±21 μg/m3. Based on the variability in elemental composition of atmospheric aerosols, using principal component analysis (PCA) and multiple linear regression (MLRA), four main sources of fine particulate matter were identified: road dust (54%), combustion of liquid fuels (4%), soil /mineral material (15%); industrial emission (10%). Also in the analysis of these data, two road dust components were identified with one being associated with local wood burning and the other with road salt.
1. INTRODUCTION
Transportation and fossil fuel combustion have been recognized to be significant and increasing sources of atmospheric particulate matter worldwide. This problem has gained additional significance as ambient particles may cause adverse health effects [1–9]. It has been suggested that fine fractions (with the diameter below 2.5 μm, PM2.5) might induce various respiratory diseases, particularly bronchitis and reduc-tion in lung funcreduc-tion parameters. These effects were observed when the mean annual concentration of particulate matter was above 20 μg/m 3 for PM2.5 or 30
μg/m 3 for
PM10 (with the diameter below 10 μm).
Wroclaw (51o07' N, 17o02' E, 116 m a.s.l.), with the total population of 634 000
inhabitants and the total area of 292.8 km2, is the fourth largest city in Poland. As
a city, it constitutes a homogeneous urban area with rural surroundings. The air quality _________________________
*Ecologistics Group, Institute of Environmental Protection Engineering, Wroclaw University of Technology, pl. Grunwaldzki 9, 50-377 Wroclaw, Poland. Corresponding author: I. Sówka, e-mail: izabela.sowka@pwr.wroc.pl
in Wroclaw is determined by a large number of emission sources, among which road traffic seems to be the most important. Traffic density in the city is very high with a dramatic frequency of severe traffic jams. Wrocław’s major industries were tradi-tionally the electronics, thermal electric power plants, and chemical factories.
The above considerations have motivated a pilot study on the PM2.5 concentration levels and elemental composition of ambient aerosol in the residential area of the city. The study of elemental composition is also essential to elucidate the sources of par-ticulate matter. Any strategy aimed at reducing pollutant levels in air demands the knowledge of chemical composition and the recognition of emission sources which significantly contribute to ambient pollutant concentrations.
2. EXPERIMENTAL
In order to assess the potential emission sources of PM2.5 in Wroclaw, a monitor-ing study was established to measure urban background concentrations in the city. In a residential area with mixed settings and allotments, PM2.5 was collected during 24 h periods from January to April, 2009.
The study employed modified IMPROVE (Interagency Monitoring of Protected Visual Environment) aerosol monitors, similar ones have been operating within the Programme IMPROVE in the United States for several years. Each of the modified IMPROVE samplers is equipped with PM2.5 Anderson inlet at approximately 3 m collecting PM2.5 on 47 mm Teflon filters (Whatman, 2 mm PTFE 46.2 mm, air flow 22.8 dm3/min).
The filter samples were analyzed at the Crocker Nuclear Laboratory, University of California at Davis (UCD). Elemental composition analysis was performed using the technique of X-ray fluorescence XRF, proton induced X-ray energy (PIXE) and pro-ton elastic scattering analysis (PESA). PESA is applicable for hydrogen determination in the sample. Total concentrations of 21 elements were analysed such as H, S, Cl, K, Ca, Ti, Fe, Mn, Cr, V, Ni, Cu, Zn, As, Pb, Br, Rb, Sr, Na, Al and Si.
Source categories for PM2.5 constituents were identified by means of the principal component analysis (PCA) using the orthogonal transformation method with varimax rotation. Multilinear regression analysis (MLRA) was applied to assess the contribu-tion of each source group to the aerosol burden [10, 11].
3. RESULTS
Table 1 summarizes the basic descriptive statistics for concentrations of PM2.5 measured at the urban background station. The PM2.5 levels varied between 12 and 100 μg/m3, with the mean value of 36±21 μg/m3. There is no the 24h PM2.5 EU target
value, only the annual one for protection of health of 25 μg/m3 (from 1.1.2010). The
World Health Organization (WHO) Air Quality Guidelines (AQG) value for 24 h PM2.5 is more restrictive and equals the annual EU target value [12]. According to the WHO statement, meeting the guideline value for 24h mean will protect against short-term high pollution that would lead to the substantial health risk.
T a b l e 1 Basic descriptive statistics calculated for 24 h PM2.5 [μg/m3] concentrations for the sampling period January/March/April 2009 Value PM2.5 (19.01–6.02) PM2.5(2.03–29.03) PM2.5 (6.04–27.04) Minimum 18 12 12 Maximum 100 85 52
Average 50 32 24
Standard deviation 24 18 12
The measured concentrations of PM2.5 were very high, particularly in Janu-ary/February, well above the AQG. The samples containing above 25 μg/m3 of PM2.5
constituted 52% (Fig. 1). The scenario may be worst because only 3 months were ana-lyzed in Wroclaw, and the highest concentrations were monitored during wintertime which is six month long or even longer.
Fig. 1. Histogram of 24 h PM2.5 concentrations for the sampling period (January/March/April 2009)
Four principal components were identified with eigenvalues higher than 1.0. The loading extent (≥ 0.7 in this study) of the correlations between the principal
compo-nents and the original variables was used to assign source identity to each one of the PCs. It can be seen from Table 2 that four components accounted for approximately 81 % of the total variance in the dataset. The identified major sources included: road dust, combustion of liquid fuels, soil/mineral material; As/Mn industrial emission.
T a b l e 2 Results of analyses of elemental components of PM2.5 aerosol in Wroclaw
from January to April 2009 (bold marked factor loadings ≥ 0.70)
Component PC1 PC2 PC3 PC4 Eigenvalues 10.53 2.99 2.27 1.22 Cum. eigenvalues 10.53 13.52 15.79 17.01 % Variance 50.14 14.22 10.82 5.82 Cum. % variance 50.14 64.36 75.18 81.00 Variable H 0.81 0.25 0.24 0.41 Al. –0.16 0.90 –0.20 –0.01 Si 0.10 0.91 0.27 0.08 S 0.32 0.49 0.62 0.35 Cl 0.91 –0.02 0.11 0.28 K 0.86 0.20 0.11 0.38 Ca 0.75 0.52 –0.11 0.05 Ti 0.47 0.70 0.06 0.08 V –0.17 –0.06 0.82 –0.19 Cr 0.36 0.11 0.28 0.74 Mn 0.41 0.30 0.026 0.74 Fe 0.51 0.20 –0.29 0.57 Ni 0.10 0.01 0.94 0.05 Cu 0.82 0.01 –0.16 0.42 Zn 0.83 –0.13 –0.074 0.44 As 0.17 –0.17 –0.15 0.86 Pb 0.61 –0.08 –0.06 0.73 Se 0.42 0.39 0.07 0.69 Br 0.70 0.04 0.19 0.61 Rb 0.70 0.22 0.43 0.24 Sr –0.035 –0.02 –0.31 –0.00
Source 1 represents a complex mixture of compounds with high concentrations of H, Cl, K, Ca, Cu, Zn, Br, Rb. Hydrogen originates mainly from organic substances (OC) and possibly ammonium sulfate (water evaporates during the analysis). Wood burning and vehicle emissions (gasoline) are both associated with OC and Ca, howev-er wood burning is also associated with K and Rb. Road dust and vehicle re-suspension are characterized by OC, Ca, Cl, Zn, Cu, Br, Road salt may be distin-guished by high levels of Cl, Na (not measured), and Ca. At low temperatures sodium
chloride mixed with calcium chloride is often used. It may also contain small amounts of K and Br. Bromine correlates greatly with Cl. The correlation coefficient r = 0.84 is calculated. Cu may be released from brake pads, Zn from tires and brake wear.
The presence of Al, Si and Ti is associated with mineral material or soil dust (fac-tor 2) while V and Ni characterize heavy oil combustion (fac(fac-tor 3). As and Mn, Cr, Pb present a high correlation with the factor 4. Metal smelting and burning of fossil fuels (including waste wood) are the major industrial processes that may contribute to con-tamination of air with these metals.
Multilinear regression analysis (MLRA) was applied to the experimental data, us-ing as dependent variables PM2.5 total mass concentrations and as independent varia-bles the principal component factor scores [10, 11]. The relative source contributions to each sample are shown in Fig. 2.
Fig. 2. Daily relative contribution of the identified sources to the PM2.5 mass for the sampling period (January/March/April 2009)
The average relative contributions of the identified sources to the PM2.5 mass for the whole sampling period, January/March/April, 2009 are given in Fig. 3. The figure shows that the average contribution of source 1 which was loosely identified as road dust was 54%. The daily contribution changed from 18% to 82% during the sampling period (Fig. 4). An interesting feature of this solution is the existence of two main components that have been identified as wood combustion products and road salts.
Source 2 represents soil with high concentrations of Al, Si and Ti, and contributes 15% to the total PM2.5 mass at this site. The average ratio of Al to Si concentrations is 0.22, very close to the typical ratio observed in soils, 0.29 [13]. Source 3 appears to
be heavy oil combustion with the presence of Ni and V mixed with S. This source contributes 4% to the total PM2.5 mass. Source 4 with high concentrations of As and Mn, Cr, Pb is assigned as industrial sources and contributes 10% to the total PM2.5 mass. An unexplained portion averaged 17% of PM2.5 mass. Some of the unexplained mass may be due to additional sources of elemental carbon, however, more work needs to be done to resolve this problem.
Fig. 3. Average relative contributions of the identified sources to the PM2.5 mass
Fig. 4. Comparison of the predicted total PM2.5 mass from MLR analysis with the measured PM2.5 mass concentrations
The test of the effectiveness of PCA together with MLRA is the comparison of the predicted PM2.5 mass versus the measured one. The results are presented in Fig. 4. The squared correlation coefficient R2 is 0.64.
4. CONCLUSIONS
The elemental composition data of the ambient aerosol collected at the residential site of Wrocław was studied to identify possible emission sources using PCA and MLRA. The results can be summarized as four sources: road dust, combustion of liq-uid fuels, soil/mineral material; As/Mn industrial emission were identified, road dust was the major source for the ambient aerosols. There is a strong evidence for a significant role of wood burning and road salt, the contribution of As emission source was observed, however, the industrial process was not recognized. It is likely that the main contributors are stationary combustion sources (boilers, furnaces, stoves, and fireplaces) in the residential and industrial use sectors. Other sources of variation in the concentration that might aid the separation of the emission sources should be incorporated, as wind direction and speed, time of year.
The compatibility of the predicted PM2.5 mass versus the measured one shows that the applied mathematical methods PCA and MLRA seem to be promising in de-termining the sources of atmospheric aerosol particles.
REFERENCES
[1] OSTRO B., BROADWIN R., GREEN S., FENG W.-Y., LIPSETT M., Environ. Health Perspect., 2006, 114, 29.
[2] JANSSEN N.A.H., SCHWARTZ J., ZANOBETTI A., SUH H., Environ. Health Perspect., 2002, 101, 43. [3] Monitoring ambient air quality for health impact assessment WHO Regional Publications, European
Series, No. 85, 1995.
[4] MOSHAMMER H., HUTTER H.P., HAUCK H., NEUBERGER M., Eur. Respir. J., 2006, 27 (6), 1138. [5] Council directive 1999/30/EC of 22 April 1999 relating to limit values for sulfur dioxide, nitrogen
dioxide and oxides of nitrogen, particulate matter and lead in ambient air, Official Journal of the Eu-ropean Communities, 1999, L 163 (29/06), 41-60.
[6] HAUCK H., BERNER A., FISCHER T., GOMISCEK B., KUNDI M., NEUBERGER M., PUXBAUM H., PREINING O., Atm. Environ., 2004, 38, 3905.
[7] STÖLZEL M., BREITNER S., CYRYS J., PITZ M., WÖLKE G., KREYLING W., HEINRICH J., WICHMANN H., PETERS A., J. Expo. Sci. Environ. Epidemiol., 2007, 17 (5), 458.
[8] PIEKARSKA K., ZACIERA M., CZARNY A., ZACZYŃSKA E., Environ. Prot. Eng., 2009, 1, 23. [9] MAZUREK CZ., ZWOŹDZIAK J., Ochr. Środ., 1994, 1 (52), 31.
[10] THURSTON G.D., SPENGER J.D., J. Clim. Apll. Meteo., 1985, 24, 1245.
[11] ALMEIDA S.M., PIO C.A., FREITAS M.C., REIS M.A., TRANCOSO M.A., Atm. Environ., 2005, 39, 3127. [12] WHO (World Health Organization) European Air quality guidelines for particulate matter, ozone,
nitrogen dioxide and sulfur dioxide. Global update 2005. Geneva 2006. [13] MASON B., Principles in Geochemistry, Wiley, New York, 1966.
